Curated data

Checking availability of months of data for states in SDC, looking at the curated data S3 bucket instead of Redshift.

Building off of AWS CLI commands such as:

aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/
# for one state:

aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20171031/content/state=TX/table=alert/projection=redshift/year=2017/

2018 New Production

Version as of 2018-07-20. First just tally which months have some content.

# Loop over states to find which ones have complete data in 2018, using new production version
avail.mo = vector()
use.states = sort(c(state.abb, c("DC", "CA1", "CA2", "CA3")))
find.months = formatC(1:12, width = 2, flag = "0")

for(i in use.states){
  cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', i ,'/table=alert/projection=redshift/year=2018/')
  mo_i <- system(cmd, intern = T)
  avail.mo_i <- substr(mo_i, start =nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
  
  avail.mo = rbind(avail.mo, c(i, find.months %in% avail.mo_i))
}

avail.mo <- as.data.frame(avail.mo)
colnames(avail.mo) = c("State", find.months)
DT::datatable(avail.mo) %>% formatStyle(2:13, background = styleEqual('TRUE', 'lightgreen'))

Now looking in the contents of each month/state, counting objects and total size of contents.

# Loop over states and months, to find which ones have complete days of data in 2018, using new production version.
# Also track object number and size of month folder. 
# Could make faster by making state/month commands pasted together and doing just one loop
avail.mo = size.mo = obj.mo = vector()

find.months = formatC(1:12, width = 2, flag = "0")
use.states = sort(c(state.abb, c("DC", "CA1", "CA2", "CA3")))
# use.states = c("AK", "SD")

statmo <- expand.grid(state = use.states, mo = find.months, stringsAsFactors = F)
statmo <- statmo[order(statmo$state),]

for(i in 1:nrow(statmo)){
  # cat(state, "\n")  i = 1; 
  mo = statmo[i, "mo"]; state = statmo[i, "state"]
    cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2018/month=', mo, "/")
    mo_i <- system(cmd, intern = T)
    avail.day_i <- substr(mo_i, start = nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
    
    
    size_cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2018/month=', mo, "/ --recursive --summarize")
    size_i <- system(size_cmd, intern = T)
    
    avail.mo = c(avail.mo, max(as.numeric(avail.day_i)))
    size.mo = c(size.mo, as.numeric(gsub("^(\\s*)+Total Size: ", "", size_i[length(size_i)])))
    obj.mo = c(obj.mo, gsub("^(\\s*)+Total Objects: ", "", size_i[length(size_i)-1]))

  
}

avail.mo[avail.mo==-Inf] = NA


df <- data.frame(State = rep(use.states, each = length(find.months)),
                 Month = rep(find.months, times = length(use.states)),
                 Days = as.numeric(avail.mo),
                 Size = size.mo,
                 Objects = as.numeric(obj.mo))

size.h = vector()

for(i in 1:length(size.mo)){
  smi = size.mo[i]
  class(smi) = 'object_size'
  size.h = c(size.h, format(smi, units = "auto"))

  }
  
df$Size.h = size.h

df$Value = paste(df$Days, df$Size.h, format(df$Objects, big.mark = ","), sep = "  ")
df$Value[is.na(df$Days)] = NA

df2 <- df %>% 
  select(State, Month, Value) %>%
  spread(key = Month, value = Value)

write.csv(df, file = paste0('Curated_Data_', Sys.Date(), '.csv'), row.names = F)

DT::datatable(df2)# %>% formatStyle(2:13, color = styleEqual('<NA>', 'grey20'))

Visual representation

library(plotly)
#ggplot(df, aes(x = Month, y = Objects, fill = State)) +geom_dotplot(binaxis = "y", stackdir='center', position = 'dodge')

df$usetext <- with(df, paste(State, "\n", Month, "\n",  format(Objects, big.mark = ",", digits = 0, scientific = F), "Objects", "\n", Size.h))

gp.focal <- ggplot(df %>% filter(State %in% c("AK", "AL", "CT", "DC", "FL", "IL", "IN", "MA", "MD", "NY", "PA", "TN", "TX", "UT", "WY")),
             aes(x = Month, 
                 y = Objects,
                 group = State,
                 text = usetext)) + theme_bw()

gp.focal2 <- gp.focal + geom_line(aes(color = State), 
                                  alpha = 0.8,
                                  size = 2) +
  theme(axis.text.x = element_text(size=7, angle=45),
        axis.text.y = element_text(size=7, angle=45)) +
  ylab("Count of Objects in S3") +
  xlab(" ")  + 
  guides(color=guide_legend(title="State"))
#gp.focal2

ggplotly(gp.focal2, tooltip = "usetext", hovermode="compare")

Same, but shown by size of month instead of number of objects.

library(plotly)
#ggplot(df, aes(x = Month, y = Objects, fill = State)) +geom_dotplot(binaxis = "y", stackdir='center', position = 'dodge')

df$usetext <- with(df, paste(State, "\n", Month, "\n",  format(Objects, big.mark = ",", digits = 0, scientific = F), "Objects", "\n", Size.h))

gp.focal <- ggplot(df %>% filter(State %in% c("AK", "AL", "CT", "DC", "FL", "IL", "IN", "MA", "MD", "NY", "PA", "TN", "TX", "UT", "WY")),
             aes(x = Month, 
                 y = Size/1000000,
                 group = State,
                 text = usetext)) + theme_bw()

gp.focal2 <- gp.focal + geom_line(aes(color = State), 
                                  alpha = 0.8,
                                  size = 2) +
  theme(axis.text.x = element_text(size=7, angle=45),
        axis.text.y = element_text(size=7, angle=45)) +
  ylab("Size (Mb) of monthly directory in S3") +
  xlab(" ")  + 
  guides(color=guide_legend(title="State"))
#gp.focal2

ggplotly(gp.focal2, tooltip = "usetext")

July data… what is being curated now?

i = "07"
for(state in use.states){
  cat(state, "\n")
    cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2018/month=', i, "/")
    mo_i <- system(cmd, intern = T)
    avail.day_i <- substr(mo_i, start = nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
    
    
    size_cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2018/month=', i, '/ --recursive --summarize --human-readable')
    (size_i <- system(size_cmd, intern = T))
    
    avail.mo = rbind(avail.mo, c(i, max(as.numeric(avail.day_i))))
    size.mo = rbind(size.mo, gsub("^(\\s*)+Total Size: ", "", size_i[length(size_i)]))
    obj.mo = rbind(obj.mo, gsub("^(\\s*)+Total Objects: ", "", size_i[length(size_i)-1]))

  
}

avail.mo[avail.mo==-Inf] = NA

2017

Now looping across all ‘states’ in the curated data:

# Loop over states to find which ones have month directories:
# can also try previous version, 20171031
avail.mo = vector()
use.states = c(state.abb, c("DC", "CA1", "CA2", "CA3"))
find.months = formatC(1:12, width = 2, flag = "0")
for(i in use.states){
  cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', i ,'/table=alert/projection=redshift/year=2017/')
  mo_i <- system(cmd, intern = T)
  avail.mo_i <- substr(mo_i, start =nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
  
  avail.mo = rbind(avail.mo, c(i, find.months %in% avail.mo_i))
}

avail.mo <- as.data.frame(avail.mo)
colnames(avail.mo) = c("State", find.months)
DT::datatable(avail.mo) %>% formatStyle(2:13, background = styleEqual('TRUE', 'lightgreen'))

# Loop over states and months, to find which ones have complete days of data in 2018, using new production version.
# Also track object number and size of month folder. 
# Could make faster by making state/month commands pasted together and doing just one loop
avail.mo = size.mo = obj.mo = vector()

find.months = formatC(1:12, width = 2, flag = "0")
use.states = sort(c(state.abb, c("DC", "CA1", "CA2", "CA3")))
# use.states = c("AK", "SD")

statmo <- expand.grid(state = use.states, mo = find.months, stringsAsFactors = F)
statmo <- statmo[order(statmo$state),]

for(i in 1:nrow(statmo)){
  # cat(state, "\n")  i = 1; 
  mo = statmo[i, "mo"]; state = statmo[i, "state"]
    cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2017/month=', mo, "/")
    mo_i <- system(cmd, intern = T)
    avail.day_i <- substr(mo_i, start = nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
    
    
    size_cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20180720/content/state=', state ,'/table=alert/projection=redshift/year=2017/month=', mo, "/ --recursive --summarize")
    size_i <- system(size_cmd, intern = T)
    
    avail.mo = c(avail.mo, max(as.numeric(avail.day_i)))
    size.mo = c(size.mo, as.numeric(gsub("^(\\s*)+Total Size: ", "", size_i[length(size_i)])))
    obj.mo = c(obj.mo, gsub("^(\\s*)+Total Objects: ", "", size_i[length(size_i)-1]))

  
}

avail.mo[avail.mo==-Inf] = NA


df <- data.frame(State = rep(use.states, each = length(find.months)),
                 Month = rep(find.months, times = length(use.states)),
                 Days = as.numeric(avail.mo),
                 Size = size.mo,
                 Objects = as.numeric(obj.mo))

size.h = vector()

for(i in 1:length(size.mo)){
  smi = size.mo[i]
  class(smi) = 'object_size'
  size.h = c(size.h, format(smi, units = "auto"))

  }
  
df$Size.h = size.h

df$Value = paste(df$Days, df$Size.h, format(df$Objects, big.mark = ","), sep = "  ")
df$Value[is.na(df$Days)] = NA

df2 <- df %>% 
  select(State, Month, Value) %>%
  spread(key = Month, value = Value)

write.csv(df, file = paste0('Curated_Data_2017_', Sys.Date(), '.csv'), row.names = F)

DT::datatable(df2)# %>% formatStyle(2:13, color = styleEqual('<NA>', 'grey20'))

library(plotly)
#ggplot(df, aes(x = Month, y = Objects, fill = State)) +geom_dotplot(binaxis = "y", stackdir='center', position = 'dodge')

df$usetext <- with(df, paste(State, "\n", Month, "\n",  format(Objects, big.mark = ",", digits = 0, scientific = F), "Objects", "\n", Size.h))

gp.focal <- ggplot(df %>% filter(State %in% c("AK", "AL", "CT", "DC", "FL", "IL", "IN", "MA", "MD", "NY", "PA", "TN", "TX", "UT", "WY")),
             aes(x = Month, 
                 y = Size/1000000,
                 group = State,
                 text = usetext)) + theme_bw()

gp.focal2 <- gp.focal + geom_line(aes(color = State), 
                                  alpha = 0.8,
                                  size = 2) +
  theme(axis.text.x = element_text(size=7, angle=45),
        axis.text.y = element_text(size=7, angle=45)) +
  ylab("Size (Mb) of monthly directory in S3") +
  xlab(" ")  + 
  guides(color=guide_legend(title="State"))
#gp.focal2

ggplotly(gp.focal2, tooltip = "usetext")

Look at the previous version of the database, from 2017-10-31:

# Loop over states to find which ones have month directories:
# can also try previous version, 20171031
avail.mo = vector()
use.states = c(state.abb, c("DC", "CA1", "CA2", "CA3"))
find.months = formatC(1:12, width = 2, flag = "0")
for(i in use.states){
  cmd = paste0('aws s3 ls s3://prod-dot-sdc-curated-911061262852-us-east-1/waze/version=20171031/content/state=', i ,'/table=alert/projection=redshift/year=2017/')
  mo_i <- system(cmd, intern = T)
  avail.mo_i <- substr(mo_i, start =nchar(mo_i[1])-2, stop = nchar(mo_i[1])-1)
  
  avail.mo = rbind(avail.mo, c(i, find.months %in% avail.mo_i))
}

avail.mo <- as.data.frame(avail.mo)
colnames(avail.mo) = c("State", find.months)
datatable(avail.mo) %>% formatStyle(2:13, background = styleEqual('TRUE', 'lightgreen'))

2018